The present invention relates, in general, to a process for fabricating lasers capable of emitting blue light, and, more particularly, to the fabrication of AlGaInN-based ridge lasers utilizing etched facet technology (EFT) for producing laser devices having a dovetail-shaped ridge cross-section.
Reflective mirrors for edge-emitting semiconductor laser diodes are typically formed at the ends of a laser cavity by mechanical cleaving of a semiconductor crystal. In general, for any semiconductor material, cleaving is an imprecise process compared to photolithography. In addition, it necessitates the handling of fragile bars or miniscule chips for device testing and other subsequent operations. It also tends to be incompatible with monolithic integration because it requires that the wafer be physically broken to obtain fully functional lasers.
Cleaving of GaN is especially problematic. Nichia Chemical first demonstrated GaN-based blue lasers on sapphire substrates in 1995 and has subsequently been able to produce commercially available CW lasers [S. Nakamura, et al. 2000 “The Blue Laser Diode: The Complete Story,” Springer-Verlag]. Cleaving is commonly used to form the facets of blue lasers, but the prices of these devices have remained very high. Cleaving the sapphire substrate to form the GaN-based laser facets is particularly difficult, since sapphire has many cleave planes with approximately equal cleave strength within a small angular distance of each other. Because of this, the fracture interface can easily be redirected from one cleavage plane to another, even when perturbations during the cleaving process are small, and when this occurs, the laser is unusable. Despite these problems, sapphire has been the substrate of choice for nitride growth because it is relatively inexpensive and stable during the high temperature processes required for GaN deposition. However, both sapphire and the more expensive SiC substrates are significantly lattice mismatched to GaN, producing high defect densities in the grown material. Free standing GaN substrates are a partial solution, and are just now becoming available, but unlike cubic InP and GaAs, GaN is hexagonal in crystal structure and much harder to cleave. It is therefore expected that cleaving will continue to be a challenging process even with GaN substrates. By using tilted substrates in CAIBE, vertical etched facet blue lasers have been fabricated [Kneissl et al., Appl. Phys. Lett. 72, 1539-1541]; however, these lasers were of the stripe or gain-guided kind. Accordingly, there is a need for an improved process for fabricating ridge-type blue lasers in a reliable and cost-effective manner.
A significant factor affecting the yield and cost of GaN-based blue lasers is the lack of availability of laser quality material with low defect density. A few research labs have developed techniques such as epitaxial lateral overgrowth (ELOG) on sapphire that have improved the defect density to the 105/cm2 level. Because of the difficulty in cleaving, described previously, the minimum cavity length that can be realistically fabricated today is on the order of 600 μm. The use of etched facets in place of mechanically cleaved facets allows the formation of shorter cavity devices of 100 μm or less. The ability to make shorter cavity devices results in a lower probability of having defects in the device and hence produces a much higher yield. These lasers may have a lower maximum power rating than longer cavity devices; however, the vast majority of lasers will be used in next generation DVD read-only applications, where lower power is sufficient and the lowest cost and lowest power consumption will be needed. The specific fabrication, integration and full wafer testing capabilities enabled by EFT will also provide significant benefits to the fabrication of high-power GaN lasers for writable optical disk applications.
Another significant problem encountered with GaN-based lasers is the provision of a good p-contact to this material. Although improvements have been made, the p-contact resistance with this material continues to be more challenging than with other III-V materials. The difficulty is that in order to get a good single lateral mode laser, the ridge for these devices needs to have a width on the order of 1 μm, and forming a good p-contact on such a small width is particularly challenging.